评估开源稀疏自动编码器在 GPT-2 Small 中析取事实知识的能力

Maheep Chaudhary, Atticus Geiger
{"title":"评估开源稀疏自动编码器在 GPT-2 Small 中析取事实知识的能力","authors":"Maheep Chaudhary, Atticus Geiger","doi":"arxiv-2409.04478","DOIUrl":null,"url":null,"abstract":"A popular new method in mechanistic interpretability is to train\nhigh-dimensional sparse autoencoders (SAEs) on neuron activations and use SAE\nfeatures as the atomic units of analysis. However, the body of evidence on\nwhether SAE feature spaces are useful for causal analysis is underdeveloped. In\nthis work, we use the RAVEL benchmark to evaluate whether SAEs trained on\nhidden representations of GPT-2 small have sets of features that separately\nmediate knowledge of which country a city is in and which continent it is in.\nWe evaluate four open-source SAEs for GPT-2 small against each other, with\nneurons serving as a baseline, and linear features learned via distributed\nalignment search (DAS) serving as a skyline. For each, we learn a binary mask\nto select features that will be patched to change the country of a city without\nchanging the continent, or vice versa. Our results show that SAEs struggle to\nreach the neuron baseline, and none come close to the DAS skyline. We release\ncode here: https://github.com/MaheepChaudhary/SAE-Ravel","PeriodicalId":501347,"journal":{"name":"arXiv - CS - Neural and Evolutionary Computing","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Evaluating Open-Source Sparse Autoencoders on Disentangling Factual Knowledge in GPT-2 Small\",\"authors\":\"Maheep Chaudhary, Atticus Geiger\",\"doi\":\"arxiv-2409.04478\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"A popular new method in mechanistic interpretability is to train\\nhigh-dimensional sparse autoencoders (SAEs) on neuron activations and use SAE\\nfeatures as the atomic units of analysis. However, the body of evidence on\\nwhether SAE feature spaces are useful for causal analysis is underdeveloped. In\\nthis work, we use the RAVEL benchmark to evaluate whether SAEs trained on\\nhidden representations of GPT-2 small have sets of features that separately\\nmediate knowledge of which country a city is in and which continent it is in.\\nWe evaluate four open-source SAEs for GPT-2 small against each other, with\\nneurons serving as a baseline, and linear features learned via distributed\\nalignment search (DAS) serving as a skyline. For each, we learn a binary mask\\nto select features that will be patched to change the country of a city without\\nchanging the continent, or vice versa. Our results show that SAEs struggle to\\nreach the neuron baseline, and none come close to the DAS skyline. We release\\ncode here: https://github.com/MaheepChaudhary/SAE-Ravel\",\"PeriodicalId\":501347,\"journal\":{\"name\":\"arXiv - CS - Neural and Evolutionary Computing\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-05\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - CS - Neural and Evolutionary Computing\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.04478\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Neural and Evolutionary Computing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.04478","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

摘要

在机理可解释性方面,一种流行的新方法是在神经元激活上训练高维稀疏自动编码器(SAE),并使用 SAE 特征作为分析的原子单位。然而,关于 SAE 特征空间是否有助于因果分析的证据尚不充分。在这项工作中,我们使用 RAVEL 基准来评估在 GPT-2 small 的隐藏表征上训练出来的 SAE 是否拥有一组特征集,可以分别传递城市在哪个国家和哪个大洲的知识。我们以神经元作为基线,以通过分布式对齐搜索(DAS)学习到的线性特征作为天际线,对 GPT-2 small 的四个开源 SAE 进行了对比评估。对于每种方法,我们都会学习一个二进制掩码,以选择将被修补的特征,从而在不改变大陆的情况下改变一个城市的国家,反之亦然。我们的结果表明,SAE 难以达到神经元基线,而且没有一个能接近 DAS 的天际线。我们在此发布代码:https://github.com/MaheepChaudhary/SAE-Ravel
本文章由计算机程序翻译,如有差异,请以英文原文为准。
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
Evaluating Open-Source Sparse Autoencoders on Disentangling Factual Knowledge in GPT-2 Small
A popular new method in mechanistic interpretability is to train high-dimensional sparse autoencoders (SAEs) on neuron activations and use SAE features as the atomic units of analysis. However, the body of evidence on whether SAE feature spaces are useful for causal analysis is underdeveloped. In this work, we use the RAVEL benchmark to evaluate whether SAEs trained on hidden representations of GPT-2 small have sets of features that separately mediate knowledge of which country a city is in and which continent it is in. We evaluate four open-source SAEs for GPT-2 small against each other, with neurons serving as a baseline, and linear features learned via distributed alignment search (DAS) serving as a skyline. For each, we learn a binary mask to select features that will be patched to change the country of a city without changing the continent, or vice versa. Our results show that SAEs struggle to reach the neuron baseline, and none come close to the DAS skyline. We release code here: https://github.com/MaheepChaudhary/SAE-Ravel
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
Hardware-Friendly Implementation of Physical Reservoir Computing with CMOS-based Time-domain Analog Spiking Neurons Self-Contrastive Forward-Forward Algorithm Bio-Inspired Mamba: Temporal Locality and Bioplausible Learning in Selective State Space Models PReLU: Yet Another Single-Layer Solution to the XOR Problem Inferno: An Extensible Framework for Spiking Neural Networks
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 联系我们:info@booksci.cn
Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。
Copyright © 2023 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1